How Much Water Does AI Use? — Personal Water Footprint Calculator

Your daily AI usage

Text queries per day

ChatGPT, Claude, Gemini, Copilot chat — count each back-and-forth exchange

AI images per day

DALL-E, Midjourney, Stable Diffusion, Adobe Firefly

Code completions per day

GitHub Copilot, Cursor, or similar AI coding suggestions accepted

Voice AI minutes per day

ChatGPT voice, Siri, Alexa, Google Assistant

— ml Per day

— liters Per week

— liters Per year

What that looks like

Water figures based on Li et al. (2023), "Making AI Less 'Thirsty'", UC Riverside. Text query: ~25 ml/exchange; image generation: ~100 ml/image; code completion: ~8 ml/suggestion; voice: ~12 ml/minute. These are server-side cooling water estimates for US data centers (WUE ~1.8 L/kWh average). Individual figures vary by data center and model.

Why does AI use so much water?

AI models run on GPU clusters that generate enormous heat. Data centers cool those servers using chilled water systems — and that water evaporates into the atmosphere, it doesn't return to local watersheds. A single large data center can consume 1–5 million gallons of water per day.

The problem is location: most hyperscale AI data centers are built in places like Arizona, Nevada, and Texas — states already under severe water stress — because land is cheap and power is available. Local water utilities in these regions have reported that large data center contracts can squeeze out allocation for residential growth, agriculture, and municipal reserves. The communities bearing the water cost rarely see economic benefits proportional to that burden — data centers are capital-intensive, largely automated facilities that employ few local workers relative to their resource draw.

Training a single large AI model can consume as much water as 700,000 liters — equivalent to manufacturing 370 cars. And that's before a single user query is answered.

Common questions about AI and water use

Water use during AI inference (running queries) is separate from the much larger water cost of training — GPT-3's training alone consumed an estimated 700,000 liters. The figures in this calculator reflect inference water use at US data centers with an average WUE of ~1.8 L/kWh.

▸ Are AI companies required to disclose how much water they use?

No US law currently requires AI companies to publicly disclose their water consumption. Most large tech companies report water use voluntarily in annual sustainability reports, but these figures are typically aggregated globally and don't break down by facility, model, or query type. The EU's Corporate Sustainability Reporting Directive (CSRD) is beginning to require more granular environmental disclosure for large companies operating in Europe. Pending US water disclosure bills at the state level are tracked in the AI legislation tracker.

▸ Which AI companies use the most water?

Microsoft, Google, and Meta operate the largest AI data center footprints and consequently use the most water globally. Microsoft's 2023 Environmental Sustainability Report disclosed a 34% year-over-year increase in water consumption as AI demand surged, reaching approximately 6.4 million cubic meters. Google and Meta have reported similar growth trajectories. OpenAI does not operate its own data centers — it runs on Microsoft Azure — so its water footprint is embedded in Microsoft's totals.

▸ Can data centers use recycled or reclaimed water for cooling?

Some data centers are beginning to use recycled or reclaimed water for cooling, but it is not yet standard practice. Google operates several facilities using 100% recycled water, and Microsoft has piloted reclaimed water cooling at some campuses. The challenge is infrastructure: reclaimed water systems require separate pipes and treatment processes that most industrial areas have not built. In water-stressed states like Arizona and Nevada, local regulators are increasingly conditioning new data center permits on recycled water commitments.

▸ How does AI water use compare to other industries?

AI data centers' water use is substantial but not yet in the same league as agriculture or manufacturing at the global scale — agriculture accounts for roughly 70% of global freshwater withdrawals, while data centers are estimated at under 1%. What makes AI data center water use notable is its concentration in already water-stressed areas, its rapid growth rate, and the fact that it is driven by consumer choices rather than food or industrial necessity. A single large data center can consume 1–5 million gallons per day, comparable to a mid-sized semiconductor fab or paper mill.

▸ What can I do to reduce my personal AI water footprint?

To reduce your personal AI water footprint, use AI tools only when they provide genuine value — avoid casual or exploratory queries that could be answered with a simple search. Choose text-based AI over image generation where possible (text queries use roughly 25 ml, image generation approximately 100 ml per image). Consider whether a local AI model running on your own hardware could handle your use case, since local inference uses no data center water. Collectively, reducing unnecessary AI demand signals to companies that efficiency matters — and reducing their operating costs gives them a business case to respond.

▸ Does using a local AI model (running on my computer) use less water?

Yes — running an AI model locally on your own hardware uses essentially no cooling water, because laptops and desktops use air cooling rather than water-cooled data center infrastructure. Tools like Ollama (which runs open models like Llama and Mistral locally) shift inference compute to your device, eliminating data center water consumption for those queries entirely. The tradeoffs are higher local electricity use, slower performance for large models, and the need for a capable device. For frequent, private AI use, local models are both more water-efficient and more private.

How Much Water Does Your AI Use?

Your daily AI usage

What that looks like

Where it comes from

Why does AI use so much water?

Common questions about AI and water use

Learn more about AI's physical footprint

Get the daily AI briefing in your inbox